LTE Cell Upgrade Ranking Using UE Session Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining LTE cell upgrades based on Performance Management counters and rigid rules do not account for granular UE-specific metrics, such as quality versus distance or UE throughput versus cell load, and fail to consider the varying benefits of cell upgrades dependent on propagation environments.
Innovation Solution
A method using user session data and a combined classification network, comprising an autoencoder and classifier, to generate a cell upgrade metric by characterizing individual UE radio environments and correlating with aggregate cell metrics and geo-spatial considerations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If PM counters and rigid rules-based decisions are used for LTE cell qualification, then the decision process is simple and standardized, but the measurement precision and ability to capture granular UE-specific metrics is insufficient
Solution Approach 1:
The system segments the cell evaluation process into multiple components: PM counter aggregation at cell level, individual UE session data collection, autoencoder-based feature extraction, and classifier-based upgrade determination. This segmentation allows granular UE-specific metrics to be analyzed separately while maintaining overall system manageability.
Solution Approach 2:
The patent introduces an autoencoder as an intermediary component that transforms raw UE session data into compressed latent representations. This intermediary layer bridges the gap between detailed UE-specific measurements and the final classification decision, enabling precise metric capture without overwhelming system complexity.
2Loss of information
If aggregate cell-level PM counters are used, then the system is easy to implement, but it cannot consider granular UE-specific metrics such as quality versus distance or UE throughput versus cell load
Solution Approach 1:
The system extracts specific UE session data from the broader cell-level PM counter framework. By taking out individual UE measurements (quality vs. distance, throughput vs. cell load) and processing them through the autoencoder, the system preserves granular information that would be lost in aggregate statistics.
Solution Approach 2:
The patent adds a new dimension to cell evaluation by incorporating individual UE session data alongside traditional aggregate PM counters. The autoencoder transforms multi-dimensional UE-specific parameters into a compressed latent space, enabling the system to consider granular characteristics without proportionally increasing processing complexity.
3Adaptability or versatility
If traditional rule-based approaches are used, then the system is computationally efficient, but it cannot provide flexible investment decisions based on continuous ranking of cells
Solution Approach 1:
The system performs preliminary action by pre-training the autoencoder on UE session data and pre-ranking cells based on their upgrade potential before actual investment decisions are made. This preliminary processing creates a continuous ranking that enables flexible decision-making without requiring complex real-time computations during the investment selection phase.
Solution Approach 2:
The patent introduces dynamics by replacing static rule-based thresholds with a dynamic classification network that continuously ranks cells based on learned patterns from UE session data. The system adapts to varying cell conditions and propagation environments, providing flexible investment priorities that reflect current network states rather than fixed rules.
Data Source
AI summary
Embodiments described herein relate to methods and apparatus for determining cell upgrades, for example whether to upgrade a Long Term Evolution (LTE) cell to Massive Multiple Input Multiple Output (M-MIMO) technology. There is provided a method for determining a cell upgrade metric. The method receives samples having user session data from user equipment communicating within a cell. Training data is obtained from the samples by applying a filtering rule and an autoencoder trained using the samples. The encoder and code layers of the trained autoencoder are combined with a classifier network to generate a combined classification network. The combined classification network is trained using the training data. The trained combined classifier network is used to classify user session data within the same or a different cell and using a plurality of the classifications to determine the cell upgrade metric for the cell.